Citation Contamination by Paper Mill Articles in Systematic Reviews of the Life Sciences
Bibliographic record
Abstract
Importance: Systematic reviews are the criterion standard for evidence synthesis in the life sciences, yet their reliability and integrity are threatened by citation contamination from fabricated publications produced by paper mills. Despite growing awareness, the extent and implications of this issue remain unclear. Objectives: To analyze the prevalence, characteristics, affected subject areas, and citation patterns of retracted paper mill articles cited in systematic reviews. Design, Setting, and Participants: This cross-sectional study analyzed systematic reviews published between 2013 and 2024, indexed in Web of Science (WoS). References were matched against the Retraction Watch dataset, and full texts were reviewed to identify retracted paper mill articles incorporated into the evidence synthesis. Main Outcomes and Measures: The study assessed (1) contamination prevalence, defined as the proportion of systematic reviews incorporating retracted paper mill articles into the evidence synthesis; (2) geographic distribution of citing authors according to institutional affiliations; (3) citation timing and trends, including the time lag between incorporation and article retraction; (4) affected research areas, categorized by WoS subject classifications; and (5) citation patterns, including highly contaminated reviews (≥3 incorporations of retracted articles). Results: Of the total of 200 000 systematic reviews, 299 incorporated at least 1 retracted paper mill article into the evidence synthesis (contamination rate, 0.15%). Among them, 256 (85.6%) included a single retracted article, and 43 (14.4%) included multiple such articles. Of 1802 author affiliations associated with the contaminated reviews, 660 (36.6%) were from institutions in China. Of 385 total citations, 124 (32.2%) occurred after retraction, including 13 occurring more than 500 days after the retraction date. Oncology was the most affected field (48 of 299 [16.1%]). Five reviews each included 5 or more retracted articles, all published in journals under questionable publishers. Conclusions and Relevance: In this cross-sectional study of life sciences systematic reviews, contamination remained low but increased over time, posing a risk to research integrity. Continued citation of retracted articles, even after retraction, highlights the need for rigorous screening practices. Correcting contaminated reviews and developing automated detection tools are essential to preserving the credibility of systematic reviews.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | MetaresearchResearch integrityBibliometrics Domain: Methods · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Observational | low |
| gpt | MetaresearchBibliometricsResearch integrity Domain: Methods · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Observational | medium |
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.258 | 0.747 |
| Meta-epidemiology (narrow) | 0.001 | 0.002 |
| Meta-epidemiology (broad) | 0.004 | 0.004 |
| Bibliometrics | 0.030 | 0.046 |
| Science and technology studies | 0.003 | 0.005 |
| Scholarly communication | 0.008 | 0.008 |
| Open science | 0.003 | 0.007 |
| Research integrity | 0.004 | 0.003 |
| Insufficient payload (model declined to judge) | 0.007 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedLabeled directly by 2 models reading the full record.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".